Morten Lauge Pedersen, Nikolai Friberg
National Environmental Research Institute, Deaprtment of Freshwater Ecology, Vejlsøvej 25, DK-8600 Silkeborg, Denmark. Tel: +45 89 20 14 00. Fax: +45 98 20 14 14. E-mail: MLP@DMU.DK
Abstract
The physical structure of two riffles in a lowland Danish stream was studied and its importance for the composition and density of the macroinvertebrate communities was evaluated. The two riffles were visually assessed to be very similar, but measurements revealed that they differed in overall hydraulic conditions, stability, substratum composition and consolidation. These differences affected abundance of both burrowing and surface dwelling macroinvertebrates. The unstable unconsolidated riffle had higher total macroinvertebrate abundance (4137 m-2 vs. 1698 m-2), diptera abundance (2329 m-2 vs. 386 m-2) and total species richness (31.7 vs. 28.8) and lower evenness (0.77 vs. 0.83) than the compact riffle. Among samples within the unconsolidated riffle, variations in macroinvertebrate communities were related to differences in mean substratum particle size. Here, a linear log-log relationship existed between macroinvertebrate abundance and the abundance of Ephemeroptera, Plecoptera and Trichoptera and the median particle size (r2total= 0.46, p=0.002; r2EPT= 0.73, p<0.001). No similar relationships were evident on the consolidated riffle. Moreover, macroinvertebrate communities on the unconsolidated riffle were dominated by species with a high colonising potential. Despite being assessed to the same geomorphological unit, inter-riffle variation was surprisingly high as the riffles differed substantially with respect to consolidation and overall hydraulic structure. These differences resulted in different macroinvertebrate community structure from the same species pool. The findings address the question if macroinvertebrate communities can be assessed at the scale of the geomorphological unit or meso-habitat.
Keywords
Streams, riffle substrata, physical habitats, spatial variation, macroinvertebrates
Introduction
Variations in catchment geology, channel morphology, discharge and sediment transport determine streambed structure and create distinct hydromorphological units such as riffles and pools within streams (Church, 1996). High-flow events exert strong physical forces that are believed to structure the overall stream morphology. The magnitude and distribution in time of these high-flow events therefore determines the overall stability of hydromorphological units. The hydro-morphological units are consequently a result of contemporary processes as well as historical hydraulic conditions.
Many lowland streams are meandering with hydromorphological units such as riffles being relatively stable over decades (Ward, 1989).
Local variations in channel morphology, sediment transport and in-stream hydraulic conditions may, however, cause significant variation in the hydromorphological units. While the overall physical structure of the stream remains constant, the finer structure of the hydromorphological units
may vary considerably due to changes in upstream morphology and local variations in hydraulics and sediment input to the stream. The habitat structure in streams is therefore the result of many physical processes acting on a number of nested spatial and temporal scales (Cummins et al., 1984; Frissell et al., 1986; Hildrew & Giller, 1994).
Riffles are normally perceived as homo-geneous hydromorphological units consisting of coarse substrata (sensu Grant et al., 1990) and their value as habitats for macroinvertebrates has been studied in comparative studies of riffles and pools (Brussock & Brown, 1991; Scarsbrook & Townsend, 1993) or in relation to sand intrusion into salmonoid spawning gravel (e.g. Sear, 1993;
Acornley & Sear, 1999; Kondolf, 2000). To our knowledge, no studies have, however, directly analysed interactions between riffle structure, sub-stratum composition and macroinvertebrate com-munities in lowland streams.
In-stream macroinvertebrate distribution is governed by the availability of different habitats and food resources and by biotic interactions (Giller & Malmquist, 1998). However, streams are
heterogeneous systems with significant variations in macroinvertebrate communities over short distances. Substratum characteristics such as particle size (e.g. Pennak & Van Gerben, 1947), stability (Stanford & Ward, 1983) and hetero-geneity (Hynes, 1970; Tolkamp, 1980) are likely to influence macroinvertebrate distribution and colonisation in the streams. Substratum texture may be another important factor (Harman, 1972;
Lamberti & Resh, 1979; Erman & Erman, 1984).
Some studies analyse the physical structure of different hydromorphological units or habitats and relate these to the macroinvertebrate community structure (e.g. Wood et al., 1999; Kemp et al., 2000) but detailed studies of specific hydromor-phological units are to our knowledge non-existent.
Our main objective was to study how physical variations in two natural riffles in a Danish lowland stream affected the macroinvertebrate community. Meandering lowland streams are thought to be relatively stable with large-scale morphology being stable over decades (Church, 1996). If so, inter-riffle variability in physical structure should be low both on the spatial and temporal scale and macroinvertebrate community structure should reflect these stable conditions.
Methods
Study sites
Two riffles placed about 100 m apart in Tange stream, Denmark were studied (Fig. 1). The two riffles are located on a 1-km reach where natural channel morphology has been preserved. The riffles were located at similar points just downstream of a meander and stream slope was 2
‰ at both riffles. Sediment transport and upstream hydraulic conditions were also identical at the two riffles. The overall morphological conditions were thus identical between the riffles and both riffles also had identical macroinvertebrate colonisation potentials. Tange stream is 4 to 6 m wide with a mean depth of 40 cm. Mean annual discharge at the site is 0.8 m3 s-1 (range 0.3 – 7.1 m3 s-1). The Tange stream catchment is characterised by a relatively quick response to precipitation events and the river thus has a peak-dominated hydrologic regime (Fig. 2). The stream flows in a valley with steep slopes, which are dominated by deciduous forest. Both riffles are partly shaded from tree canopies extending from the stream bank. Overall channel morphology is characterised by natural irregular meanders and the stream is morphologically active and capable of migrating freely within the river valley.
Tange stream Norway
Sweden
Germany Denmark
N
0 100 500 m
n = 9
n = 9
Downstream riffle Upstream riffle
Tange Stream
Figure 1. Location of the field sites and sampling methodology on the riffles in Tange stream. The insert shows sampled fields in the two transects (grey colour).
Discharge (m3 s-1)
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002
0 1 2 3 4 5 6 8
7
Figure 2. Daily mean discharge time series from Tange stream 1990-2001.
Riffle characteristics
Sampling was carried out during a period of low flow on two consecutive days in May 2001. Ten transects were evenly spaced along each entire riffle. Each transect was divided into plots, each 0.5 m wide and extending 1 m upstream from the transect line. In the centre of the plot, water depth was measured to nearest cm and current velocity was measured 5 cm above the streambed with an inductive current meter (OTT, Nautilus C2000). In each plot the substrate distribution was determined using the following categories: Stones (> 64 mm), coarse gravel (8-64 mm), fine gravel (2-8 mm), coarse sand (1-2 mm) and fine sand (0.1-1 mm). The presence of hard clay, peat and mud (<0.1 mm black) was recorded along with the coverage of organic debris such as dead plant tissue and leaves. Data were used for an overall characterisation of the two riffles. A total of 105 plots were sampled on the upstream riffle and 122 on the downstream riffle.
The discharge was measured using a propeller current meter (OTT instruments, Germany). Ten current-velocity profiles were measured across the stream downstream of the studied reach using a propeller current meter.
Each velocity profile represented 1/10 of the stream width. The discharge was calculated by integrating the velocity profiles over the depth and multiplying by the width. The slope was calculated from optical levelling of the stream bed (Levelling instrument: Zeiss Instruments, Germany).
To assess substratum stability of the riffles, 50 stones of equal size (approx. 35 mm x 60 mm) painted with yellow dye on the upper side were placed in 5 clusters in different parts of the riffles
at different velocities. The stones were placed a month prior to the experiment and were examined and removed on the first day of sampling. Each stone was given a score in the field prior to removal from 0 to 4 based on the dye removal as a consequence of scouring. The following scoring system was used: (0) no visible scouring; (1): 0-25%
of the painted surface scoured; (2): 25-50% of the painted surface scoured; (3) more than 50% of the dye removed by scouring; (4) turned – dye-side faced down.
As an additional measure of riffle surface stability, the algal biomass was used. The painted stones were transported to the laboratory in opaque plastic containers and 96% ethanol was added to the plastic containers until the stones were covered. All samples were extracted at 5ºC and kept in darkness for 12 h following 15 min of ultrasound treatment. Thereafter they were filtered through a GF/C filter and the volume of ethanol used for each stone was measured. The chlorophyll content was determined spectro-photometrically as described by Søndergaard &
Riemann (1979).
Riffle consolidation was measured by means of a penetrometer (Sear, 1995). The resistance to penetration is a function of the density of sediment packing and the degree of interlock. By applying a consistent force 3 times to an iron rod (length: 50 cm) and measuring the penetration depth a semi-quantitative measure of consolidation is achieved. Penetrometer measure-ments were carried out in 36 points on each riffle.
Physical habitat and macroinvertebrate sampling Two transects were randomly selected for the intensive study of physical habitats and macro-invertebrates. Each transect was divided into 4 quadrates (1 m x 1 m) evenly distributed across the wetted width. Each quadrate was further sub-divided into 4 fields (0.5 m x 0.5 m). Nine fields out of 16 were sampled in each transect using a nested randomised design, sampling 3 out of 4 fields in 3 of 4 quadrates (Fig. 1). In each selected field, sampling depth and current velocity 5 cm above the bed were measured in the each corner and in the centre of the field. Macroinvertebrates were sampled using a 200 cm2 surber-sampler (200 µm mesh size). The streambed was disturbed and all material to a depth of 5 cm was retained in the mesh bag. Subsequently, the material from each sample was transferred to a container, preserved in 70% ethanol and transported to the laboratory where all macroinvertebrates were identified to either species or genus level, except for dipterans that were identified to sub-family level.
The organic fraction was separated from the sample, and CPOM (>1 mm) and FPOM (<1 mm) were separated by sieving. The organic
fractions were dried for at least 6 hours at 60° C until constant weight. After dry combustion at 550°C for 1 hour the ash-free dry weight (AFDW) was calculated for each fraction. The inorganic substratum was then wet-sieved through a series of stainless steel sieves (diameter = 20 cm, Endecotts, London). The following fractions were dried and weighed: 64 mm, 32 mm, 16 mm, 8 mm, 4 mm, 2 mm, 1 mm, 0.5 mm and <0.5 mm. Particle size distribution and median particle size were calculated for each sample.
Data analyses and statistical methods
The spatial distribution of water depth and current velocity was calculated and a 2D-plot for each riffle was generated using a kriging method in the Surfer v. 7.0 software package (Golden Software, 1999). Mean values and standard deviation of depth and current velocity between riffles were compared using standard t-tests on log-transformed data to satisfy assumptions of normality and equal variances (Snedecor &
Cochran, 1989). The reach-scale distributions between riffles were compared using Kolomogorov-Smirnoff tests (Conover, 1980).
The substratum distribution based on the survey of the entire riffle was calculated and compared using a Kolomogorov-Smirnoff test and the median particle size was compared between riffles using a t-test. To assess the stability of the riffles, the scouring-scores for each riffle were compared using a Kolomogorov-Smirnoff test. A Spearman rank correlation analysis was performed on physical habitat variables (Conover, 1980).
Macroinvertebrate community structure and diversity were expressed in several ways.
Mean abundance and abundance of Ephemerop-tera, Plecoptera and Trichoptera (EPT), Shannon-Wiener diversity (H’) and Pielou’s evenness (J’) were calculated for each riffle (Washington, 1984).
Taxonomic richness was also calculated as the mean number of taxa (S) on each riffle and the overall species richness (Smax) estimated from the 1st order Jack-knife estimate based on re-sampling of the species lists (Palmer, 1990). Confidence intervals for Smax were calculated from Smith & van Belle (1984). The abundance of dipterans and their proportion in each sample were also calculated.
Samples were classified for each riffle separately using two-way indicator species analysis (TWINSPAN). TWINSPAN analysis was carried out on presence-absence data in the PC-ORD Version 4.23 software package (McCune &
Mefford, 1999). The TWINSPAN groups were tested for significance calculating Bray-Curtis similarities between samples and testing these among the TWINSPAN groups. All statistical tests were performed in the SAS system version 8.2 (SAS Institute, 2000). Differences in environmental
variables (depth, current velocity and median particle size) were tested among the TWINSPAN groups by means of ANOVA. Pair-wise differences in environmental variables were tested using standard t-tests on log-transformed data.
Results
Overall characterisation of riffle habitats
Discharge at the study sites was 0.6 m3 s-1 which is close to mean summer discharge. The length of the upstream- and downstream riffle was 18 m and 10 m, respectively. Mean depth and current velocity were significantly higher (22 cm and 44 cm s-1) on the downstream than on the upstream riffle (19 cm
and 34 cm s-1). Median particle size in the 18 macroinvertebrate samples was not significantly different between the riffles (Table 1; t-test, p>0.05).
0 10 20 30 40 50 60 70
0 5 10 15 20 25 30 35 40 45
1 m 1 m Sand Fine gravel Stone + coarse gravel
Downstream riffle Upstream riffle
Depth (cm) Velocity (cm s-1) Substratum
Depth Velocity Substratum
Figure 3. Contour plots of depth (cm), near bed current velocity (cm s-1) and substrata on the upstream and downstream riffle.
Variation in habitat structure
The physical habitat structure on the riffles, as assessed by contour plots of depth, current velocity and substratum, demonstrated large differences (Fig. 3). On the upstream riffle the majority of the flow (approx. 80%) was concentrated in a relatively deep channel with high current velocity in one side of the riffle, while the other shallow part had low current velocity (Fig. 3). On the downstream riffle the flow pattern was more heterogeneous with high velocities and shallow and deep areas irregularly distributed throughout the riffle (Fig. 3). On the upstream riffle, the distribution of substrata showed a similar pattern, stones and gravel were located in the shallow part of the riffle with low current velocity, whereas the sandy substrata were found in the deep and fast flowing part of the riffle. On the downstream riffle the substrata were more mixed and there was an irregular pattern in the substratum composition (Fig. 3).
The spatial differences in physical riffle structure between the two sites were also reflected in the frequency distribution of depth and current velocity (Fig. 4). There was a high frequency of shallow depths (< 10 cm) on the upstream riffle, whereas depths close to the mean value (20 –30 cm) dominated the downstream riffle. Velocities of 20-40 cm s-1 dominated on the upstream riffle, whereas about 50% of all velocities on the
downstream riffle exceeded 40 cm s-1 (Fig. 4). The distribution of depth and current velocities were significantly different between the two riffles (Kolomogorov-Smirnoff test, p<0.001).
The substratum distribution was signi-ficantly different between the two riffles (Fig. 4;
Kolomogorov-Smirnoff test, p<0.05). Fine gravel was the dominant substratum on the upstream riffle (49%), whereas fine gravel (35%) and fine sand (34%) were equally dominant on the downstream riffle (Fig. 4). The fine sand cover was higher and the fine gravel coverage was lower on the downstream riffle than on the upstream riffle.
All other substrata were not significantly different between the riffles (Fig. 4).
Current velocity and the presence of coarse substrata (stones and gravel) was positively correlated among sampling fields on the downstream riffle (r = 0.39, p < 0.001). On the upstream riffle this correlation was negative (r= 0.18, p = 0.07). On both riffles, however, the highest velocities were found in areas with high coverage of fine gravel and coarse sand (rupstream = 0.54, p < 0.001; rdownstream = 0.64, p < 0.001).
The spatial distribution of depth and current velocity on the riffles demonstrated that depth and current velocity was significantly negatively correlated on the downstream riffle, whereas a slightly positive correlation existed on the upstream riffle (Fig. 5).
0-10 10-20 20-30 30-40 40-50 50-60
Percentage (%)
Depth (cm)
Upstream Downstream
0-10 10-20 20-30 30-40 40-50 50-60 60-70 70-80 80-90 Current velocity (cm s-1)
Stones C. gravel F. gravel C. sand F. sand 0
10 20 30 40 50 60 70
Coverage (%)
0 10 20 30 40 50
0
Percentage (%)
B A
C
10 20 30 40 50
Figure 4. Depth (A), current velocity (B) and substratum distributions (C) on the two riffles in Tange stream.
Riffle substratum stability and consolidation Riffle stability was expressed as the scour of 50 stones placed in each riffle a month prior to sampling. On the downstream riffle, a significantly larger percentage of stones had been turned over (30%) than on the upstream riffle (10%) (Fig. 5;
Kolomogorov-Smirnoff test, p<0.001). Stones with no scouring constituted 44% of all stones on the upstream riffle, whereas only 4% were left without any scour on the downstream riffle (Fig. 6). On 20 naturally embedded stones of equal size on each riffle the algal biomass was significantly lower on the downstream riffle (65 mg chl. a m-2) than on the upstream riffle (105 mg chl. a m-2; t-test, p<0.001).
No scour <25% 25-50% >50% Turned 0
5 10 15 20
No. of stones
Surface scour
Upstream Downstream
Figure 6. Riffle stability measured as the percentage of stone surface impacted by scour on the two riffles.
Riffle consolidation varied significantly between the two riffles (Table 2; t-test, p<0.001).
Consolidation measured as the penetration depth was approximately 2.5 times higher at the downstream riffle than at the upstream riffle.
Macroinvertebrate communities
A total of 31 macroinvertebrate species were found on the two riffles. Species richness was 26 for the upstream riffle and 27 for the downstream riffle.
Mean species richness per sample was identical on the two riffles, but total estimated species richness was significantly higher (31.7) on the downstream
riffle than on the upstream riffle (28.8) (Table 3) as seen from the confidence limits. Mean macroinvertebrate abundance was much higher on the downstream riffle than on the upstream riffle (Table 3; t-test, p < 0.05). This difference was primarily due to a 6-fold higher diptera abundance (t-test, p < 0.05). The community diversity and the diversity and abundance of EPT species on the two riffles were identical (Table 3). Pielou’s evenness was, however, significantly higher (0.83) on the upstream riffle than on the downstream riffle (0.77), thus reflecting the lower abundance on the upstream riffle and the dominance of few abundant species on the downstream riffle (Table 3; t-test, p < 0.05).
The mayfly Baetis spp. dominated the macroinvertebrate community on the upstream riffle (31%). The second most abundant species was Gammarus pulex L. (24%), followed by the dipteran predator Dicranota spp. (11%) and chironomids (9%). The stoneflies Amphinemura standfussi Ris and Leuctra spp. together made up 10% of the macroinvertebrate community. On the downstream riffle chironomids were the most abundant macroinvertebrates (29%). The second most common species were Dicranota spp. (25%) and G. pulex (15%) followed by A. standfussi and Leuctra spp. (13%). Baetis spp. constituted only 5%
of the macroinvertebrate community on the downstream riffle.
Correlations between physical variables and macroinvertebrate community variables were restricted to correlation between median sub-stratum particle size and EPT species richness, EPT abundance, species richness and total abundance in the sample (Fig. 7). Both overall abundance and EPT abundance were correlated linearly to the median particle size in the sample on the downstream riffle (Fig. 7BD). No significant linear correlation existed on the upstream riffle (Fig.
7AC).
R2 = 0.12, p = 0.161 R2 = 0.41, p = 0.004 B
Current velocity (cm s-1) Current velocity (cm s-1) A
Depth (cm)
0 10 20 30 40 50
0 10 20 30 40 50 60 70
Depth (cm)
0 10 20 30 40 50
0 10 20 30 40 50 60 70
Figure 5. Relationship between depth and current velocity on the upstream (A) and downstream riffle (B).
TWINSPAN classification of macroinvertebrate samples
The macroinvertebrate TWINSPAN classification (presence-absence data) separated samples into 3 groups on both riffles (Fig. 8; ANOVA, p<0.05). On the upstream riffle a group (III) of 3 samples with low species richness and without Leuctra spp. and Naididae was separated in the first division (Fig.
8). The other two groups were characterised by either high species richness (group I, n = 7) or moderate species richness (group II, n = 8).
Macroinvertebrates indicator species in group I included Elmis aenea (Müller) and Heptagenia spp.
On the downstream riffle, 6 samples with high species richness were separated in the first division (group III, Fig. 8). Simulids and Baetis spp. were indicator species for this group. The remaining 12 samples were divided into a group with a core-community of the most common species (group II, n = 8) and a group (n = 4) with low species richness and low EPT taxa richness (Table 4 and Fig. 8).
Macroinvertebrate community structure in the TWINSPAN groups on the upstream riffle was only different with respect to number of individuals, evenness and EPT taxa richness.
TWINSPAN group (I) had higher abundance, lower evenness and higher EPT taxa richness than the other two groups. TWINSPAN group (II) and (III) had similar macroinvertebrate community
structures (Table 4). Macroinvertebrate community group (III) was significantly different from group (I) and (II) communities on the downstream riffle.
Species richness, macroinvertebrate abundance and Shannon diversity were higher in TWINSPAN group (III) than in group (I).
Upstream riffle
Group III 3 samples
Group II 7 samples Group I
8 samples
Downstream riffle
Group III 6 samples
Group II 8 samples Group I
4 samples
Figure 8. TWINSPAN analyses on the upstream and downstream riffle.
R2 = 0.46, p = 0.002 B
A
2.5 3.0 3.5 4.0 4.5
Log (EPT abundance m-2)
Log (median particle size)
Log (EPT abundance m-2)
Log (median particle size)
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
2.5 3.0 3.5 4.0 4.5
R2 = 0.73, p = 0.001
D C
2.5 3.0 3.5 4.0 4.5
Log (total abundance m-2)
Log (median particle size)
Log (total abundance m-2)
Log (median particle size)
0 0.5 1.0 1.5 2.0 0 0.5 1.0 1.5 2.0
2.5 3.0 3.5 4.0 4.5
Figure 7. Relationship between median particle size and total macroinvertebrate abundance and between median particle size and EPT taxa abundance on the upstream (A, C) and downstream (B, D) riffle.
TWINSPAN group I
TWINSPAN group II
TWINSPAN group III 0
10 20 30 40 50 60
Coverage (%)
0 10 20 30 40 50 60 70
Coverage (%)
Stones C. gravel F. gravel C. sand F. sand 0
10 20 30 40 50 60
Coverage (%)
Upstream
TWINSPAN group I
TWINSPAN group II
TWINSPAN group III 0
10 20 30 40 50 60
Coverage (%)
0 10 20 30 40 50 60 70
Coverage (%)
Stones C. gravel F. gravel C. sand F. sand 0
10 20 30 40 50 60
Coverage (%)
Downstream
Figure 9. Substratum distribution in the TWINSPAN groups on the riffles.
The samples were not grouped systematically in relation to hydraulic conditions or amount of POM on either riffle. Thus, the current velocity and the amount of POM in the sediment were not significantly different among TWINSPAN groups. However, on both riffles the substratum distribution varied among TWIN-SPAN groups. On the upstream riffle, the substratum distribution in group (II) was different from the distribution in group (I) and (III), and on the downstream riffle, all three distributions were significantly different (Fig. 9; Kolomogorov-Smirnoff test, p<0.05). No further physical differences were apparent on the upstream riffle.
In contrast large differences in in-stream habitat parameters were present on the downstream riffle.
The primary difference among the groups was related to the substratum distribution and the coarseness of the substratum. The median particle size was significantly higher in TWINSPAN group (III) than in TWINSPAN groups (I) and (II). (Table 5). The group (II) samples were taken in significantly deeper areas of the riffle than the samples in the third group (III). The range in physical habitat parameters was generally wider on the downstream riffle than on the upstream riffle, thus reflecting the diverse physical habitat structure on this riffle.
Discussion
Overall physical structure in streams
The two riffles in Tange stream were located at identical points in the stream and the discharge regime, sediment transport through the riffles and slope varied little between the riffles. Nonetheless, the physical structure on the two riffles varied considerably due to differences created by local
variations in streambed morphology, structure and hydraulic conditions within the individual riffle.
Local variations in in-stream physical structure We mapped near-bed flow and depth along the two riffles and could therefore demonstrate a significant variation the overall hydraulic structure between them. The upstream riffle had the majority of the flow concentrated in one part of the riffle. In contrast, the downstream riffle had the flow distributed across the entire riffle in an irregular pattern. Under low-flow conditions, flow lines in riffles diverge and this irregular flow pattern is therefore how the flow structure is normally perceived in lowland riffles (Church, 1996).
On the upstream riffle the presence of coarse substrata was maintained in areas of low current velocity due to higher velocities at other seasons. As discharge increases during the winter the flow is no longer constrained to one side of the riffle. The entire riffle will be engaged in the flow, and high current velocities over the coarse substrata will remove deposited fine material (Carling, 1996). On the downstream riffle current velocities will increase across the entire riffle as discharge increases, resulting in higher stress on the streambed across the riffle. The riffle structure is thus controlled by previous historic flow events, but local differences in stream bed sediment structure and morphology on the two riffles have caused significantly different reactions to these events, with different contemporary flow patterns and parameter relations as a result (Schumm, 1977;
Church, 1996).
Riffle stability and substratum structure
The hydraulic conditions on the riffles were reflected in the distribution of substrata across the riffles. On the upstream riffle, sand dominated in the fast flowing section of the riffle. Gravel and stones had either been eroded from this part or buried beneath sand. The sandy streambed deposits thus represent a temporary deposition as it is in transport along the streambed. The downstream riffle had a more complex spatial distribution of substrata, thus reflecting the variation in riffle hydraulic structure. Substratum composition was noticeably finer and more sand was present, reflecting the higher velocities capable of moving coarser material. The patchy environment on this riffle creates patches of lower current velocities and shear stress where different particle sizes remain stable on the streambed.
However, the overall current velocities are highest on this riffle and 44% of the stones were turned, indicating an unstable environment. The sand dominance on the streambed therefore reflects a temporary deposition, as sand is dominant in
transport along the streambed (Brookes, 1988;
Thompson, 1986).
The lower coverage of coarse substrata and the dominance of fine material on the downstream riffle indicate a lower stability here (Mangelsdorf et al., 1990). Low stability was observed by in situ use of stone clusters, which have proved a reliable and flexible method for estimating surface stability in streams (Matthaei et al., 1999; Ferguson et al., 2002). On the downstream riffle, stones in all five stone clusters were turned over. In contrast, only stones in two clusters were turned over on the upstream riffle. The unstable conditions on the downstream riffle were further suggested by the low concentration of chlorophyll a on stable natural stones in the riffle, indicating either significant scour from sediment in transport or more frequent movements of the stones (Giller
& Malmqvist, 1998). However, as macroinverte-brate abundance was higher on the downstream riffle, the low algal biomass can also be result of higher grazer pressure. This possibility cannot be excluded since the grazer Elmis aenea and the chironomid family Orhocladiinae, which includes a number of grazer species were more abundant on the downstream riffle (Merrit & Cummins, 1996).
The upstream riffle sediment structure was very compact. In contrast, sediments were loosely structured on the downstream riffle.
Reliable and robust methods for evaluating the structure of the stream bed sediments from field methods measuring in situ structure and texture are few (Cummins, 1964). Therefore we used a simple one-dimensional method that we believe is applicable in lowland streams where the stream bed sediments are heterogeneous and consists of a number of different particle fractions. Median particle size in the two riffles was identical. The difference in surface sediment size was thus not reflected in the sub-surface sediments when sampling to a depth of 5 cm. This indicates a substantial coverage of sandy substrata in the sub-surface layers on both riffles. The compact sub-surface structure on the upstream riffle indicate that this riffle had a pavement of coarse surface stones arranged in a compact pattern, whereas this pavement was irregular at the downstream riffle, thereby opening up a larger part of the sub-surface sediments for colonisation of macroinvertebrates.
The compact sediment structure in one side of the upstream riffle has affected the local streambed morphology and caused alteration of the flow pattern. This has thus enhanced stabilisation by concentrating the physical stress to a confined flow channel. The hydraulic regime in Tange stream is peak-dominated and the streambed is therefore exposed to high shear stress all year. Even during summer the discharge can
increase two-fold. These high-flow events can potentially cause instability within the stream given the right unconsolidated conditions. The downstream riffle is in a transitional unstable phase with fast flow and fine substrata, despite the riffle-pool sequences usually being considered stable over decades (Frissell et al., 1986; Ward, 1989).
Macroinvertebrate communities on riffles
Differences in macroinvertebrate communities were caused by a significantly higher abundance of chironomids and Dicranota spp. on the down-stream than the updown-stream riffle. The uncon-solidated nature of the downstream riffle enhances the possibility of colonisation by the multi-voltine, burrowing r-strategists such as Dicranota spp., and most chironomids (Merritt & Cummins, 1996).
Dicranota spp. is a predator and its high abundance indicates favourable habitat conditions and high prey density. The flux through the riffle of prey for Dicranota spp. and fine particulate organic matter (FPOM) for the chironomids (Tanytarsini and some Orthocladiinae) need to be high to keep the relative abundance at 25% and 29% of all individuals for the two taxa. Gammarus pulex and Baetis spp. were abundant on the upstream riffle, indicating the existence of a widespread sheltered zone of low flow (Wiberg-Larsen, 1984; Dahl &
Greenberg, 1996). Amphinemura standfussi and Leuctra spp. are usually found in areas of relatively stable substrata (Wiberg-Larsen, 1984). The presence of these taxa on the downstream riffle reflects that sheltered refuge areas are present on parts of the riffle, despite the low stability (Lancaster & Hildrew, 1993). Only part of the riffle surface is presumably moving and an intermediate stability regime is perhaps present here (Stanford
& Ward, 1983).
Interactions between macroinvertebrates and substratum are well documented and many studies have focused on relating macroinvertebrate species composition and abundance to substratum types, flow and depth (e.g. Brunke et al., 2001).
Most of these studies have concentrated on measuring colonisation on substrata of different particle sizes and on uniform and mixed substrata (Pennak & Van Gerben, 1947; Ward, 1975). The results from these studies show that species richness, diversity and macroinvertebrate abundance increase with particle size. We found identical substrata on both studied riffles and a substantial species overlap was observed among samples and between riffles.
Variation in macroinvertebrate abundance was linearly related to median particle size on the unconsolidated downstream riffle. No linear correlation existed on the upstream riffle, indicating that abundance on the consolidated